From: "Nautiyal, Ankit K" <ankit.k.nautiyal@intel.com>
To: "Ville Syrjälä" <ville.syrjala@linux.intel.com>
Cc: <intel-gfx@lists.freedesktop.org>, <intel-xe@lists.freedesktop.org>
Subject: Re: [PATCH 5/5] drm/i915/vrr: Move the TGL SCL manging of vmin/vmax/flipline deeper
Date: Thu, 18 Sep 2025 19:12:40 +0530 [thread overview]
Message-ID: <b29061f7-230e-43be-8e2d-1c3d51752f63@intel.com> (raw)
In-Reply-To: <aMv3uG3NnKPSpJE4@intel.com>
[-- Attachment #1: Type: text/plain, Size: 9691 bytes --]
On 9/18/2025 5:44 PM, Ville Syrjälä wrote:
> On Thu, Sep 18, 2025 at 03:03:02PM +0530, Nautiyal, Ankit K wrote:
>> On 9/18/2025 2:04 AM, Ville Syrjala wrote:
>>> From: Ville Syrjälä<ville.syrjala@linux.intel.com>
>>>
>>> Currently our crtc_state->vrr.{vmin.vmax,flipline} are mangled on
>>> TGL to account for the SCL delay (the hardware requires this mangling
>>> or the actual vtotals will become incorrect).
>> Please correct me if I am wrong:
>>
>> For display < 13 VRR hardware inserts one extra scanline just after
>> vactive. So we need to have flipline, vmax, vmin vtotal shifted by 1.
>> This is computed in intel_vrr_extra_vblank_delay()
> vmin/vmax/flipline stay put, but we need to account for that extra
> scanline when calculating the guadband/pipeline_full.
>
>> For display < 13 hardware expects flipline >= vmin (so we add flipline +
>> 1, not related to vblank delay.
>>
>> For some platforms intel_psr_min_vblank_delay() of 1 is required so that
>> moves the vblank start with this amount. we can account for this delay
>> by intel_vrr_real_vblank_delay().
> That 1 scaline is just the minimum size for SCL. And once we start
> tracking the SCL explicitly then AFAICS pretty much all
> intel_vrr_real_vblank_delay() calls can just be replaced with
> crtc_state->set_context_latency.
>
>> So for the SCL delay mentioned here should'nt we
>> use intel_vrr_extra_vblank_delay() in the computation below?
> I think we should be able to get rid of intel_vrr_extra_vblank_delay()
> for the most part. As state we just need to account for it when
> calculating the guardband/pipeline_full.
Hmm do you mean something like below:
void intel_vrr_compute_scl(struct intel_crtc_state *crtc_state)
{
struct intel_display *display = to_intel_display(crtc_state);
int scl = 0; /*lines before delayed vblank */
if (!HAS_DSB(display))
return 0;
/*Account for 1 scanline thing for ICL/TGL */
scl = intel_vrr_extra_vblank_delay(display);
scl = max(scl, intel_psr_min_vblank_delay(crtc_state));
crtc_state->vrr.set_context_latency = scl;
}
And then in place of intel_vrr_real_vblank_delay() we can use the
crtc_state->vrr.set_context_latency.
Regards,
Ankit
>
>>
>> Regards,
>>
>> Ankit
>>
>>
>>
>>> Unfortunately this
>>> means that one can't simply use these values directly in many places,
>>> and instead we always have to go through functions that undo the
>>> damage first. This is all rather fragile.
>>>
>>> Simplify our lives a bit by hiding this mangling deeper inside
>>> the low level VRR code, leaving the number stored in the crtc
>>> state actually something that humans can use.
>>>
>>> This does introduce a dependdency as intel_vrr_get_config()
>>> will now need to know the SCL value, which is read out in
>>> intel_get_transcoder_timings(). I suppose we could simply
>>> duplicate the SCL readout in both places should this become
>>> a real hinderance. For now just leave a note around the
>>> intel_get_transcoder_timings() call to remind us.
>>>
>>> Signed-off-by: Ville Syrjälä<ville.syrjala@linux.intel.com>
>>> ---
>>> drivers/gpu/drm/i915/display/intel_display.c | 4 ++
>>> drivers/gpu/drm/i915/display/intel_vrr.c | 76 +++++++++++---------
>>> 2 files changed, 47 insertions(+), 33 deletions(-)
>>>
>>> diff --git a/drivers/gpu/drm/i915/display/intel_display.c b/drivers/gpu/drm/i915/display/intel_display.c
>>> index f4124c79bc83..18b9baa96241 100644
>>> --- a/drivers/gpu/drm/i915/display/intel_display.c
>>> +++ b/drivers/gpu/drm/i915/display/intel_display.c
>>> @@ -3901,6 +3901,10 @@ static bool hsw_get_pipe_config(struct intel_crtc *crtc,
>>> pipe_config->framestart_delay = 1;
>>> }
>>>
>>> + /*
>>> + * intel_vrr_get_config() depends on TRANS_SET_CONTEXT_LATENCY
>>> + * readout done by intel_get_transcoder_timings().
>>> + */
>>> if (!transcoder_is_dsi(pipe_config->cpu_transcoder) ||
>>> DISPLAY_VER(display) >= 11)
>>> intel_get_transcoder_timings(crtc, pipe_config);
>>> diff --git a/drivers/gpu/drm/i915/display/intel_vrr.c b/drivers/gpu/drm/i915/display/intel_vrr.c
>>> index 71fb64c92165..71a985d00604 100644
>>> --- a/drivers/gpu/drm/i915/display/intel_vrr.c
>>> +++ b/drivers/gpu/drm/i915/display/intel_vrr.c
>>> @@ -156,25 +156,13 @@ static int intel_vrr_vblank_exit_length(const struct intel_crtc_state *crtc_stat
>>>
>>> int intel_vrr_vmin_vtotal(const struct intel_crtc_state *crtc_state)
>>> {
>>> - struct intel_display *display = to_intel_display(crtc_state);
>>> -
>>> /* Min vblank actually determined by flipline */
>>> - if (DISPLAY_VER(display) >= 13)
>>> - return intel_vrr_vmin_flipline(crtc_state);
>>> - else
>>> - return intel_vrr_vmin_flipline(crtc_state) +
>>> - intel_vrr_real_vblank_delay(crtc_state);
>>> + return intel_vrr_vmin_flipline(crtc_state);
>>> }
>>>
>>> int intel_vrr_vmax_vtotal(const struct intel_crtc_state *crtc_state)
>>> {
>>> - struct intel_display *display = to_intel_display(crtc_state);
>>> -
>>> - if (DISPLAY_VER(display) >= 13)
>>> - return crtc_state->vrr.vmax;
>>> - else
>>> - return crtc_state->vrr.vmax +
>>> - intel_vrr_real_vblank_delay(crtc_state);
>>> + return crtc_state->vrr.vmax;
>>> }
>>>
>>> int intel_vrr_vmin_vblank_start(const struct intel_crtc_state *crtc_state)
>>> @@ -258,6 +246,21 @@ void intel_vrr_compute_vrr_timings(struct intel_crtc_state *crtc_state)
>>> crtc_state->mode_flags |= I915_MODE_FLAG_VRR;
>>> }
>>>
>>> +static int intel_vrr_hw_value(const struct intel_crtc_state *crtc_state,
>>> + int value)
>>> +{
>>> + struct intel_display *display = to_intel_display(crtc_state);
>>> +
>>> + /*
>>> + * On TGL vmin/vmax/flipline also need to be
>>> + * adjusted by the SCL to maintain correct vtotals.
>>> + */
>>> + if (DISPLAY_VER(display) >= 13)
>>> + return value;
>>> + else
>>> + return value - intel_vrr_real_vblank_delay(crtc_state);
>>> +}
>>> +
>>> /*
>>> * For fixed refresh rate mode Vmin, Vmax and Flipline all are set to
>>> * Vtotal value.
>>> @@ -265,14 +268,7 @@ void intel_vrr_compute_vrr_timings(struct intel_crtc_state *crtc_state)
>>> static
>>> int intel_vrr_fixed_rr_hw_vtotal(const struct intel_crtc_state *crtc_state)
>>> {
>>> - struct intel_display *display = to_intel_display(crtc_state);
>>> - int crtc_vtotal = crtc_state->hw.adjusted_mode.crtc_vtotal;
>>> -
>>> - if (DISPLAY_VER(display) >= 13)
>>> - return crtc_vtotal;
>>> - else
>>> - return crtc_vtotal -
>>> - intel_vrr_real_vblank_delay(crtc_state);
>>> + return intel_vrr_hw_value(crtc_state, crtc_state->hw.adjusted_mode.crtc_vtotal);
>>> }
>>>
>>> static
>>> @@ -441,14 +437,6 @@ void intel_vrr_compute_config_late(struct intel_crtc_state *crtc_state)
>>> crtc_state->vrr.pipeline_full =
>>> intel_vrr_guardband_to_pipeline_full(crtc_state,
>>> crtc_state->vrr.guardband);
>>> -
>>> - /*
>>> - * vmin/vmax/flipline also need to be adjusted by
>>> - * the vblank delay to maintain correct vtotals.
>>> - */
>>> - crtc_state->vrr.vmin -= intel_vrr_real_vblank_delay(crtc_state);
>>> - crtc_state->vrr.vmax -= intel_vrr_real_vblank_delay(crtc_state);
>>> - crtc_state->vrr.flipline -= intel_vrr_real_vblank_delay(crtc_state);
>>> }
>>> }
>>>
>>> @@ -607,6 +595,21 @@ void intel_vrr_set_db_point_and_transmission_line(const struct intel_crtc_state
>>> EMP_AS_SDP_DB_TL(crtc_state->vrr.vsync_start));
>>> }
>>>
>>> +static int intel_vrr_hw_vmin(const struct intel_crtc_state *crtc_state)
>>> +{
>>> + return intel_vrr_hw_value(crtc_state, crtc_state->vrr.vmin);
>>> +}
>>> +
>>> +static int intel_vrr_hw_vmax(const struct intel_crtc_state *crtc_state)
>>> +{
>>> + return intel_vrr_hw_value(crtc_state, crtc_state->vrr.vmax);
>>> +}
>>> +
>>> +static int intel_vrr_hw_flipline(const struct intel_crtc_state *crtc_state)
>>> +{
>>> + return intel_vrr_hw_value(crtc_state, crtc_state->vrr.flipline);
>>> +}
>>> +
>>> void intel_vrr_enable(const struct intel_crtc_state *crtc_state)
>>> {
>>> struct intel_display *display = to_intel_display(crtc_state);
>>> @@ -616,11 +619,11 @@ void intel_vrr_enable(const struct intel_crtc_state *crtc_state)
>>> return;
>>>
>>> intel_de_write(display, TRANS_VRR_VMIN(display, cpu_transcoder),
>>> - crtc_state->vrr.vmin - 1);
>>> + intel_vrr_hw_vmin(crtc_state) - 1);
>>> intel_de_write(display, TRANS_VRR_VMAX(display, cpu_transcoder),
>>> - crtc_state->vrr.vmax - 1);
>>> + intel_vrr_hw_vmax(crtc_state) - 1);
>>> intel_de_write(display, TRANS_VRR_FLIPLINE(display, cpu_transcoder),
>>> - crtc_state->vrr.flipline - 1);
>>> + intel_vrr_hw_flipline(crtc_state) - 1);
>>>
>>> intel_de_write(display, TRANS_PUSH(display, cpu_transcoder),
>>> TRANS_PUSH_EN);
>>> @@ -754,6 +757,13 @@ void intel_vrr_get_config(struct intel_crtc_state *crtc_state)
>>> crtc_state->vrr.vmin = intel_de_read(display,
>>> TRANS_VRR_VMIN(display, cpu_transcoder)) + 1;
>>>
>>> + if (DISPLAY_VER(display) < 13) {
>>> + /* undo what intel_vrr_hw_value() does when writing the values */
>>> + crtc_state->vrr.flipline += intel_vrr_real_vblank_delay(crtc_state);
>>> + crtc_state->vrr.vmax += intel_vrr_real_vblank_delay(crtc_state);
>>> + crtc_state->vrr.vmin += intel_vrr_real_vblank_delay(crtc_state);
>>> + }
>>> +
>>> /*
>>> * For platforms that always use VRR Timing Generator, the VTOTAL.Vtotal
>>> * bits are not filled. Since for these platforms TRAN_VMIN is always
[-- Attachment #2: Type: text/html, Size: 11005 bytes --]
next prev parent reply other threads:[~2025-09-18 13:42 UTC|newest]
Thread overview: 21+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-09-17 20:34 [PATCH 0/5] drm/i915/vrr: Hide icl/tgl idiosyncrasies better Ville Syrjala
2025-09-17 20:34 ` [PATCH 1/5] drm/i915/vrr: Extract helpers to convert between guardband and pipeline_full values Ville Syrjala
2025-09-18 9:39 ` Nautiyal, Ankit K
2025-09-17 20:34 ` [PATCH 2/5] drm/i915/vrr: Readout framestart_delay earlier Ville Syrjala
2025-09-18 9:40 ` Nautiyal, Ankit K
2025-09-17 20:34 ` [PATCH 3/5] drm/i915/vrr: Store guardband in crtc state even for icl/tgl Ville Syrjala
2025-09-18 9:37 ` Nautiyal, Ankit K
2025-09-18 12:21 ` Ville Syrjälä
2025-09-18 13:59 ` Nautiyal, Ankit K
2025-09-18 18:00 ` Nautiyal, Ankit K
2025-09-17 20:34 ` [PATCH 4/5] drm/i915/vrr: Annotate some functions with "hw" Ville Syrjala
2025-09-18 9:41 ` Nautiyal, Ankit K
2025-09-17 20:34 ` [PATCH 5/5] drm/i915/vrr: Move the TGL SCL manging of vmin/vmax/flipline deeper Ville Syrjala
2025-09-18 9:33 ` Nautiyal, Ankit K
2025-09-18 12:14 ` Ville Syrjälä
2025-09-18 13:42 ` Nautiyal, Ankit K [this message]
2025-09-18 14:18 ` Ville Syrjälä
2025-09-18 17:58 ` Nautiyal, Ankit K
2025-09-17 20:48 ` ✓ CI.KUnit: success for drm/i915/vrr: Hide icl/tgl idiosyncrasies better Patchwork
2025-09-17 21:28 ` ✓ Xe.CI.BAT: " Patchwork
2025-09-18 2:13 ` ✗ Xe.CI.Full: failure " Patchwork
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=b29061f7-230e-43be-8e2d-1c3d51752f63@intel.com \
--to=ankit.k.nautiyal@intel.com \
--cc=intel-gfx@lists.freedesktop.org \
--cc=intel-xe@lists.freedesktop.org \
--cc=ville.syrjala@linux.intel.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox